I have been researching using Stata commands for propensity score matching and
sensitivity analyses in relation to selection bias. I have already used the
command -pscore- from Stata Journal. Now that I have a propensity score, I
would like to conduct a sensitivity analysis. In my research, I came across
-psmatch2- (from ssc), -rbounds-(from ssc), and -mhbounds- (from Stata
Journal). I have read through the help files and searched Statalist for uses;
however, I want to make sure that my use is appropriate.
I am currently using Stata 9.2 to start to get used to the commands, but I will
be conducting the final analysis on a computer with Stata 10. My project
includes examining selection bias into private schools and the effects school
sector on test
scores and taking trigonometry. The following is my syntax
-psmatch2 private, outcome(testscore) pscore(mypscore)-
I did not use common support with the -pscore- command, so I did not include
that in my -psmatch2- syntax. When I have run the
-psmatch2-command, Stata warns me that:
"There are observations with identical propensity score values.
The sort order of the data could affect your results.
Make sure that the sort order is random before calling psmatch2."
Since I am not using -psmatch2- to create the propensity score, do I need to be
concerned about the sorting? If so, is there a way to make the order of my data
random?
I also plan to use -rbounds- and -mhbounds-, with the following syntax
-gen delta=testscore-_testscore if _treat==1-
(do I need to create a delta for _treat==0?)
-rbounds delta, gamma (1 (.05)2)-
-mhbounds trig, gamma (1 (.05)2)-
Does it appear that I am correctly using these commands?
Thank you for your consideration,
Elizabeth Covay
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
